He Sold His Startup to Twitter. Now He’s Building AI That Codes | Tomas Halgas
In this episode of The Everyday Founder, James sits down with Tomas Helgaš, founder and CEO of Sutro — the AI platform that turns a single text prompt into a full production-ready app.
Before Sutro, Tomas worked at Facebook (helping power the “People You May Know” feature), then went on to build and sell Sphere to Twitter after raising $30M.
Now, he’s back for round two — building Sutro to change the way the world builds software. This conversation dives into the mindset, mission, and mechanics behind one of the most fascinating builders in AI today.
We discuss:
- Why Tomas left his dream job at Facebook to start Sphere
- How he built a company that became the foundation for X (Twitter) Communities
- The early days of Sutro and building secure, production-grade AI tools
- The tradeoff between shipping fast vs. building safely
- Co-founders, hiring mistakes, and second-time founder lessons
- How AI will reshape how software is built — and why security is the next big frontier
Chapters:
00:00 – Mission: change how the world builds software
01:18 – Intro: Tomas Halgaš, Sutro, Facebook → Sphere → Twitter
02:08 – Why go back after a successful exit
03:10 – Early bet on OpenAI (GPT-2) and text-to-app ideas
04:02 – Viral demos: prompt → live front- & back-end, fully deployed
05:12 – Sphere origin story & community product
07:42 – From partnership talks to Twitter acquisition
09:24 – Post-acquisition & starting Sutro (overlap period)
11:05 – Co-founder transition; going solo as CEO
12:21 – Sutro thesis: production software with guarantees
14:02 – Security horror stories & why prototypes don’t cut it
15:42 – Comp landscape: “shipping fast” vs “building safely”
16:57 – The mission (again): security-first software at scale
18:13 – Sponsor: Opus
18:38 – From prompt links to better human–computer interaction
19:13 – Product/LLM inflection points; scaffolds & reasoning
21:13 – Moving upmarket: serious, enterprise use cases
21:39 – Raising capital: when VC makes sense
23:16 – What a “good VC” actually does for founders
28:15 – Second-time founder perspective on investor value
28:42 – Family roots, teenage hacking, first sparks
30:16 – Silicon Valley internship & mindset shift
31:01 – Leaving Facebook: parental push & perspective
33:02 – Hiring mistakes at Sphere: brilliant but hard to work with
34:12 – New bar at Sutro: great humans, ex-founders
35:04 – Remote vs in-person; hiring uncommon talent
37:41 – Advice to younger self: read deeply, then build deeply
41:01 – Working with people: persuasion vs. coaching
42:03 – Coaching someone with a “dream job” to leap
43:18 – Calibrate at the best; then go build
45:21 – CTO vs CEO: different learning curves
46:04 – Big tech cycles & urgency vs comfort
47:03 – Co-founders: marriage, commitment, alignment
49:13 – The “F-you number” & exit alignment
51:21 – Host anecdote; aligning on outcomes
52:19 – Post-acq reality: agency & decision-making changes
54:21 – Corporate planning constraints & politics
55:24 – How to set better terms for future acquisitions
56:10 – Where Sutro is going
57:16 – Better models, but HCI is the hard problem
59:16 – English is ambiguous: need precise interfaces
1:00:19 – The compliance/security wave is coming
1:01:27 – Sutro’s advantage: guarantees, compliance, real software
1:01:54 – The next interface: text + visuals + flows
1:02:17 – AI moves fast; strategy expires quickly
1:02:39 – Luck vs talent vs hard work
1:03:39 – Where to follow Tomas & try Sutro
1:04:03 – Outro
Hosted on Acast. See acast.com/privacy for more information.
